You are on page 1of 7

Basics of EXPERIMENT

DISADVANTAGES
Experimentation
- Lab findings are artificial +
EXPERIMENTAL RESEARCH tendency to generalize real life
situations
● Technique that focuses on - Lab approach is time consuming
cause-effect relationships + difficult to design
● Enables researchers to identify - Experimental approach is not
causal relationships as it leads to fully adequate to make it valid
the observation under controlled and reliable measure of human
conditions of the effects of behavior because the later is too
systematically changing one or complex
more variables through precise
instruments / tools WHY EXPERIMENTS ARE
CONDUCTED?
Psychological Experiments: objective
analysis & observations of phenomena
- Test theories & provide database
made to occur under controlled
for explanation of behavior
situations.
- See what happens in the absence
- Objective = free from biases
of a compelling theory [requires
no knowledge of theory or
EXPERIMENT ADVANTAGES existing database]
- Repeat / replicate previous
- Identifies the strength of casual findings [extend previous
relationships. procedure by adding something
- Inferential strength is derived new while retaining something
from the degree of control that old]
can be applied
- One / more variables can be under
the control of researcher /
experimenter
- Findings are applicable to
everyday life situations /
problems.
- [Differences among subjects,
Research Variables
equipment failures, inconsistent
instructions]
INDEPENDENT VARIABLE - Confounding: happens when a
[IV] value of EV changes
- Stimulus variable systematically across different
- Antecedent condition deliberately conditions of an experiment
manipulated to assess its effects - Experimental results
on behavior cannot be interpreted with
- The condition the experimenter certainty + causal
chooses to vary relationships cannot be
- Its values are set by the inferred
experimenter, and not affected by - Make sure no EV change along
anything else that happens in the the IV
experiment
- There must be at least 2 levels of RESEARCH PROBLEM
IV [music: classical & rock]
STATEMENT OF THE PROBLEM
DEPENDENT VARIABLE [DV] - Description of an issue currently
- Response variable existing w/c needs to be addressed
- Dependent measures - Provides context for the research
- Depends on the stimulus given study + generates questions which
- Particular behavior expected to the research aims to answer
change because of the - Focal point of any research
experimental treatment - [The frequency of job layoffs is
- Outcome we are trying to explain creating fear, anxiety, and a loss
- Its values are assumed to depend of productivity in middle
on the values of IV management workers]
- Change in IV = Change in DV
Key Components: often have 3 elements
EXTRANEOUS VARIABLE - Problem itself, stated clearly and
- Factors that are not the focus of w/ enough contextual detail to
experiment, but can influence the establish importance
findings - Method of solving the problem,
- Neither intentionally manipulated often stated as a claim / working
IV / DV measured as indexes of thesis
the effect of the IV
- Purpose, statement of objective, [intends to explore +
and scope of the project being describe behaviors as they
proposed naturally occur]

Criteria ● Experimental: tentative


- Clearly indicate what is to be explanation of an event / behavior
investigated - Statement that explains the
- Actual statement may be in a effects of specified
declarative or in question form antecedent conditions on a
- Indicate the variables of interest + measured behavior
specific relationship between
variables to be studied Formulating Hypothesis: CRITERIA

FORMULATING HYPOTHESIS ● Synthetic statements: can either


be true / false [can be supported
HYPOTHESIS or contradicted]
- Thesis / main idea of an ○ Hungry students read
experiment slowly
- Statement about a predicted ○ Avoid non-synthetic
relationship between at least two statements [analytical /
variables contradictory]
- Non-scientific synonyms: ○ Analytical: always true
speculation, guess, hunch ○ Contradictory: always
- Statement of a research false. statements oppose
hypothesis is designed to fit the each other.
type of research design selected ○ We don’t need to conduct
an experiment because we
NON EXPERIMENTAL VS already know what the
outcome will be
EXPERIMENTAL
○ Can be stated in “if-then”
HYPOTHESIS
form

● Nonexperimental: statement of
● Testable statements: means for
predictions of how events, trait, or
manipulating antecedent condition
behaviors might be related - not a
& measuring the resulting
statement about cause-effect
behavior must exist
- Some do not typically
include a hypothesis
● Falsifiable statements:
disprovable by the research INTUITION: knowing w/o reasoning
findings - Not necessarily unscientific =
inferences drawn from intuition
● Parsimonious statements: simplest can sometimes violate scientific
explanation criteria
○ Not so many supporting - Most accurate if it comes from
assumptions experts; good hunches are really
an unconscious result of our own
● Fruitful statements: leads to new expertise in an area
studies
Operational Definitions
4 MODELS: Come Up with a
Hypothesis - Specifies the precise meaning of a
variable in terms of observable
INDUCTIVE MODEL: process of operations, procedures,
reasoning from specific cases to more measurements.
general principles - Operational: clearly describes
- Often used in science and math operations involved in
- Certain specifics first to reach a manipulating variables.
more general conclusion - Statements of operating
procedures, sets of instructions
DEDUCTIVE MODEL: reverse of that tell others how to carry out an
inductive experiment.
- From general principles to make
predictions about specifics OPERATIONAL DEFINITIONS vs
instances DICTIONARY DEFINITIONS
- Most useful with a
well-developed theory with “Anxiety”
clearly stated basic premises
Dictionary definition: state of concern or
SERENDIPITY: knack of findings that painful uneasiness about a future or
are not being sought uncertain event.
- Ivan Pavlov’s classical
conditioning - Adequate for everyday use but not
- Matter of knowing enough to use in the context of experiment
an opportunity because it does not tell how to
produce different levels / values ● “Hungry condition”: should
of the variables. consistently show signs of hunger
[increased food-seeking behavior]
- Don’t have procedures to follow
to make people anxious or - Several sets of measurements
non-anxious. according to O.D. of D.V., we
should get the same results each
OPERATIONAL DEFINITION time
- Use standardized test that have
EXPERIMENTAL O.D. shown to be reliable
- independent variable in detail
- Describes what exactly was done HOW TO CHECK RELIABILITY?
to create various treatment
conditions of the experiment Interrater Reliability
- Includes all steps followed to set
up each value of the I.V. - Have varying observers that take
measurements of the same
MEASURED O.D. responses
- dependent variable; what we do to - Agreement between their
measure measurements
- Exact procedures followed to - Can be statistically compared
assess the impact of different [reliability coefficients]
treatment conditions - From 0.0 [only chance levels of
- Exact descriptions of specific agreement] to 1.0 [perfect
behaviors / responses recorded + agreement]
explanation how those responses
are scored Test-Retest Reliability

- Comparing scores of people who


EVALUATING O.D.
have been measuring twice with
the same instrument
RELIABILITY: consistency &
- They take the test once, and take
dependability
it again [after a reasonable
interval]
● If applied in more than one
experiment, they ought to work in
similar ways each time [EOD]
Interim Reliability - Manipulation check: provide
evidence for validity of
- Different parts of a questionnaire, experimental procedure
test, or other instruments are
designed to assess the same HOW TO ESTABLISH EVIDENCE
variable attain consistent results FOR VALIDITY?
[internal consistency]
Face Validity
● Split-half reliability: Involves
splitting test into two halves at - Variables that can be manipulated
random & computing coefficient and measured fairly directly
reliability between scores - Least stringent type of validity =
obtained on two halves does not provide any real
evidence
- Two halves should strongly - Validity of MOD centers on the
correlate if items are question of whether we measure
measuring the same what we intend to measure
variable
- Effects of pupil size = we simply
● Cronbach’s alpha [a]: statistical use a ruler
test evaluating internal
consistency of entire set of items Content Validity

- Most widely used method - Degree to w/c content of a


for evaluating interim measure reflects the content of
reliability what is being measured
- Considers the correlation
of each test item with every - Does the content of our measure
other item fairly reflect the content of the
quality we are measuring?
VALIDITY: actually studying the
variable we intend to study - Are all aspects of the content
- If researchers actually measure represented properly?
what they intend to measure
- Necessary to provide evidence for
the validity of experimental
manipulation
Predictive Validity

- A measuring instrument yields


info allowing prediction of actual
behavior / performance

- Do our procedures yield info that


enables us to prejudice future
behavior? - they should if we are
measuring what we intend to
measure

Concurrent Validity

- Degree to w/c scores on


measuring instrument correlate
with another known standard for
measuring the variable being
studied

Construct validity

- An O.D. accurately represents the


construct it is intended to
manipulate
- Deals with transition from theory
to research application
- Start with the general idea of the
qualities that characterize the
construct we want to test
- We seek to find ways of putting
our idea to an empirical test

You might also like